Objective To assess the causal relationship between coffee intake and prostate cancer risk by using the two-sample Mendel randomization (MR) method. Methods The genome-wide association study (GWAS) data on coffee intake (exposure) and prostate cancer (outcome) were obtained from two independent data sets in UK Biobank. The inverse variance weighted method (IVW), weighted median estimator method (WME), and MR-Egger method were used for MR analyses. The OR value and 95%CI were used to represent the association between coffee intake and prostate cancer. In addition, the MR-Egger method was performed for pleiotropic and heterogeneity tests, and the leave-one-out method was used for sensitivity analysis. Results A total of 38 SNP were selected as instrumental variables. The IVW method showed that coffee intake might reduce the risk of prostate cancer (OR=0.994; 95%CI: 0.990-0.999; P=0.009). The WME method obtained the same conclusions (OR=0.991; 95%CI: 0.985-0.999; P=0.018), but MR-Egger regression did not find a causal relationship between coffee intake and prostate cancer (OR=0.992; 95%CI: 0.983-1.000; P=0.084). The MR-Egger method showed no pleiotropy (intercept=4.2E-5; P=0.581) or heterogeneity (Q=27.20; P=0.854) among the instrumental variables. The sensitivity analysis indicated that the conclusion was robust. Conclusion Two-sample Mendel randomization analysis reveals that coffee consumption might reduce the risk of prostate cancer.